Neurotoxins, such as those obtained from Clostridium botulinum and Clostridium tetani, are highly potent and specific poisons of neural cells, and other cells when delivered within such cells for therapeutic purposes. These Gram positive bacteria express two related but distinct toxins types, each comprising two disulfide-linked amino acid chains: a light chain (L) of about 50 KDa and a heavy chain (H) of about 100 KDa, which are wholly responsible for the symptoms of these diseases. The holotoxin is synthesised in vivo as a single-chain, then nicked in a post-translational modification to form the active neurotoxin comprising the separate L and H chains.
The tetanus and botulinum toxins are among the most lethal substances known to man, having a lethal dose in humans of between 0.1 ng and 1 ng per kilogram of body weight. Tonello et al., Adv. Exp. Med. & Biol. 389:251-260 (1996). Both toxins function by inhibiting neurotransmitter release in affected neurons. The tetanus neurotoxin (TeNT) acts mainly in the central nervous system, while botulinum neurotoxin (BoNT) acts at the neuromuscular junction and other cholinergic synapses in the peripheral nervous system; both act by inhibiting neurotransmitter release from the axon of the affected neuron into the synapse, resulting in paralysis.
The tetanus neurotoxin (TeNT) is known to exist in one immunologically distinct type; the botulinum neurotoxins (BoNT) are known to occur in seven different immunogenic types, termed BoNT/A through BoNT/G. While all of these types are produced by isolates of C. botulinum, two other species, C. baratii and C. butyricum also produce toxins similar to /F and /E, respectively. See e.g., Coffield et al., The Site and Mechanism of Action of Botulinum Neurotoxin in Therapy with Botulinum Toxin 3-13 (Jankovic J. & Hallett M. eds. 1994), the disclosure of which is incorporated herein by reference.
Regardless of type, the molecular mechanism of intoxication appears to be similar. In the first step of the process, the toxin binds to the presynaptic membrane of the target neuron through a specific interaction between the heavy (H) chain and a cell surface receptor; the receptor is thought to be different for each type of botulinum toxin and for TeNT. Dolly et al., Seminars in Neuroscience 6:149-158 (1994), incorporated by reference herein. The carboxyl terminus of the heavy chain appears to be important for targeting of the toxin to the cell surface. Id.
In the second step, the toxin crosses the plasma membrane of the poisoned cell. The toxin is first engulfed by the cell through receptor-mediated endocytosis, and an endosome containing the toxin is formed. The toxin then escapes the endosome into the cytoplasm of the cell. This last step is thought to be mediated by the amino terminus of the H chain, which triggers a conformational change of the toxin in response to a pH of about 5.5 or lower. Endosomes are known to possess a proton pump which decreases intra endosomal pH. The conformational shift exposes hydrophobic residues in the toxin, which permits the toxin to embed itself in the endosomal membrane. The toxin then translocates through the endosomal membrane into the cytosol.
The last step of the mechanism of botulinum toxin activity appears to involve reduction of the disulfide bond joining the H and light (L) chain. The entire toxic activity of botulinum and tetanus toxins is contained in the L chain of the holotoxin; the L chain is a zinc (Zn++) endopeptidase which selectively cleaves proteins essential for recognition and docking of neurotransmitter-containing vesicles with the cytoplasmic surface of the plasma membrane, and fusion of the vesicles with the plasma membrane. TxNT, BoNT/B BoNT/D, BoNT/F, and BoNT/G cause degradation of synaptobrevin, also called vesicle-associated membrane protein (VAMP), a synaptosomal membrane protein. Most of the cytosolic domain of VAMP extending from the surface of the synaptic vesicle is removed as a result of any one of these cleavage events. Each toxin (except TeNT and BoNT/B) specifically cleaves a different bond.
BoNT/A and /E selectively cleave the plasma membrane-associated protein SNAP-25; this protein, which is also cleaved by BoNT/C1 (Foran et al., Biochem. 35:2630-2636 (1996)), is predominantly bound to and present on the cytosolic surface of the plasma membrane. BoNT/C cleaves syntaxin, an integral protein having most of its mass exposed to the cytosol. Syntaxin interacts with the calcium channels at presynaptic terminal active zones. See Tonello et al., Tetanus and Botulinum Neurotoxins in Intracellular Protein Catabolism 251-260 (Suzuki K. & Bond J. eds. 1996), the disclosure of which is incorporated by reference as part of this specification.
Both TeNT and BoNT are taken up at the neuromuscular junction. BoNT remains within peripheral neurons, and blocks release of the neurotransmitter acetylcholine from these cells. Through its receptor, TeNT enters vesicles that move in a retrograde manner along the axon to the soma, and is discharged into the intersynaptic space between motor neurons and the inhibitory neurons of the spinal cord. At this point, TeNT binds receptors of the inhibitory neurons, is again internalized, and the light chain enters the cytosol to block the release of the inhibitory neurotransmitters 4-aminobutyric acid (GABA) and glycine from these cells.
Because of its specifically localized effects, minute doses of BoNT have been used since 1981 as therapeutic agents in the treatment of patients suffering from dystonias, including strabismus (misalignment of the eye), bephlarospasm (involuntary eyelid closure) and hemifacial spasm. See e.g., Borodic et al, Pharmacology and Histology of the Therapeutic Application of Botulinum Toxin in Therapy with Botulinum Toxin 119-157 (Jankovic J. & Hallett eds. 1994), hereby incorporated by reference herein. Of the seven toxin types, BoNT/A is the most potent of the BoNTs, and the best characterized. Intramuscular injection of spastic tissue with small quantities of BoNT/A has also been used effectively to treat spasticity due to brain injury, spinal cord injury, stroke, multiple sclerosis and cerebral palsy. The extent of paralysis depends on both the dose and volume delivered to the target site.
Although the L chain is the moiety responsible for neural intoxication, it must be delivered to the neural cytoplasm in order to be toxic. Similarly, the single-chain holotoxin pro-forms exhibit relatively low toxicity until they are cleaved at one or more peptide bonds in an exposed loop region between their H and L chains to create the fully-active mature neurotoxins. As implied in the mechanism provided above, the H chain of each neurotoxin is essential for cell receptor binding and endocytosis, while both the L and the H chains (and an intact disulfide bond) are required for translocation of the toxin into the cytoplasm. As indicated above, the L chain alone is responsible for the toxicity caused by inhibition of acetylcholine secretion.
Despite the clear therapeutic efficacy of clostridial neurotoxin preparations, industrial production of the toxin is difficult. Production of neurotoxin from anaerobic Clostridium cultures is a cumbersome and time-consuming process including a multi-step purification protocol involving several protein precipitation steps and either prolonged and repeated crystallisation of the toxin or several stages of column chromatography. Significantly, the high toxicity of the product dictates that the procedure must be performed under strict containment (BL-3). During the fermentation process, the folded single-chain neurotoxins are activated by endogenous clostridial proteases through a process termed nicking. This involves the removal of approximately 10 amino acid residues from the single-chain to create the di-chain form in which the two chains remain covalently linked through the interchain disulfide bond.
The nicked neurotoxin is much more active than the unnicked form. The amount and precise location of nicking varies with the serotypes of the bacteria producing the toxin. The differences in single-chain neurotoxin activation and, hence, the yield of nicked toxin, are due to variations in the type and amounts of proteolytic activity produced by a given strain. For example, greater than 99% of C. botulinum type A single-chain neurotoxin is activated by the Hall A C. botulinum strain, whereas type B and E strains produce toxins with lower amounts of activation (0 to 75% depending upon the fermentation time). Thus, the high toxicity of the mature neurotoxin plays a major part in the commercial manufacture of neurotoxins as therapeutic agents.
The degree of activation of engineered clostridial toxins is, therefore, an important consideration for manufacture of these materials. It would be a major advantage if neurotoxins such as BoNT and TeNT could be expressed in high yield in rapidly-growing bacteria (such as heterologous E. coli cells) as relatively non-toxic single-chains (or single-chains having reduced toxic activity) which are safe, easy to isolate and simple to convert to the fully-active form.
With safety being a prime concern, previous work has concentrated on the expression in E. coli and purification of individual H and L chains of TeNT and BoNT; these isolated chains are, by themselves, non-toxic; see Li et al., Biochemistry 33:7014-7020 (1994); Zhou et al., Biochemistry 34:15175-15181 (1995), hereby incorporated by reference herein. Following the separate production of these peptide chains and under strictly controlled conditions the H and L subunits can be combined by oxidative disulphide linkage to form the neuroparalytic di-chains. Unfortunately, this strategy has several drawbacks.
Firstly, it is not practical to express and isolate large amounts of the individual chains; in particular, in the absence of the L chain the isolated H chain is quite insoluble in aqueous solution and is highly susceptible to proteolytic degradation. Secondly, the in vitro oxidation of the individually expressed and purified H and L chains to produce the active di-chain is very inefficient, and leads to low yields of active toxin and the production of many inactive incorrectly folded or oxidized forms. The purification of the correctly folded and oxidized H and L chain-containing toxin is difficult, as is its separation from these inactive forms and the unreacted separate H and L chains.
It would therefore be useful and advantageous to express clostridial neurotoxins as inactive (or less active) single-chain forms, to eliminate the need for the time-consuming and inefficient reconstitution of the constituent chains, to maintain solubility of the protein chains, to reduce protein misfolding and consequent susceptibility to protease attack, to improve toxin yield, and/or to provide a simple method for the purification of the toxin.
Additionally, it would be useful to engineer these toxins to provide single-chain, modified neurotoxin molecules having novel therapeutic properties and/or longer duration of action, or toxic or non-toxic forms for use as transport molecules capable of delivering a therapeutic moiety to nerve or other cell types. By expressing such proteins as a single-chain, the yield and purification of the engineered proteins would be vastly improved.